The 1D Average pooling block represents an average pooling operation. This block outputs a smaller tensor than its input, which means downstream blocks in your model will need fewer parameters and amount of computation; it also serves to control overfitting.
The 1D Average pooling block moves a pool (window) with a set size over the incoming data, computing the average in each specific window. How big steps the window takes is determined by the stride.
Average pooling blocks are inserted after one or more convolutional blocks; they help inner convolutional block receive information from a bigger portion of the original vector. If we see convolutional blocks as detectors of a specific feature, average pooling finds the “mean” value of that feature inside the pooling vector. Each channel (hence each feature) is treated separately.
Size: The length of the vector with which the average is computed.
Stride: Distance between the left edge of consecutive pooling windows.
Padding: Same results in padding the input such that the output has the same length as the original input. Valid means "no padding".
Stay in the know by signing up for occasional emails with tips, tricks, deep learning insights, product updates, event news and webinar invitations.
We promise not to spam you or share your email with any third party. You can change your preferences at any time. See our privacy policies.
Please check your email inbox account to confirm, set, or update your communication preferences.